Strange color points

[CTZn]

Right, reinhard tonemapping will detect the disruption of a single atom if there is no other light source. Only Camera and linear tonemapping represent a constant sensitivity to light, they must be used prioritarily (?) when setting lights up !

Last night I was updating the Maya section a bit in wiki, I'll go further today. I swear the next MtI version is out in 36h max

[neepneep]

This happens to me 100% of the time when I have a material that uses dispersion (cauchy > 0) If dispersion is disabled then everything looks fine.

Can anyone confirm that this is a bug or if I am just doing something wrong?

[PureSpider]

How high do you set your cauchy?

[neepneep]

oh my mistake, its not due to cauchy - its something else...

Try run the dispersion test scene and see for yourself. Every time I try it I get loads of multicoloured fireflies. (try with aperture diffraction disabled)


Next edit the igs file and change the cauchy to 0 then run it. The fireflies are still there but only in 2 colours.

[PureSpider]

The dispersion test scene always looked like that, with that colored splotches.
Don't ask me why though...

[neepneep]

It seems that each indigo version previous to 0.9_test1 will render the scene slightly different but looks correct - , while versions 0.9_test1 and onwards renders with those dots - and it gets more intense it later versions.

So which version is correct? Or is there still a bug here? (or just operator error?)

[Zom-B]

very nice testrange neepneep, even with indigo 0.4 ^^

Now lets wait for some words from the master

[CTZn]

I'm betting on a bidir effect, early Indigo versions were mlt based if I am correct. Switch to mlt only and you will have similar results, or I'm a fool (huh, I'm taking some risks here).

[suvakas]

Yep, CTZn is correct.
Here's the MLT version renered with 1.1.15

[neepneep]

ah, excellent thanks fellas, I guess that solves it - it's been awhile and I completely forgot about those bidir issues

edit: with a 10,000 spp render but not enough rainbow

[Sinister]

I was away for some time..I swiched to new indigo and MTI. It seems that dots dissapear both with reinhrd tonemapping and linear/camera tonemapping. Still some dust, but I'm not shure that it mti or indigo issue

[Kram1032]

neepneep: Try a higher cauchy b, then

[CTZn]

Sinister: in the last release I fixed (reverted oops) both camera aperture and efficacy scale, maybe that helped if you used the meshlight settings I posted and corrected the glass scene scale

Was bidir on for your test ? By the way are you talking about the dispersion test scene or your glass scene ?

[Sinister]

Bi Dir = on, Adv Opt = on,3100spp (best result, but color dust in shadow)


Bi Dir = on, Adv Opt = off , 3500spp


Bi Dir = off, Adv Opt = on, 2900spp (Disco )


Bi Dir = off, Adv Opt = off, 3500 spp
(looks just the same as second picture (BiDir is a part of Adv Opt?) )

[CTZn]

It's nice to have done these tests Sinister, but maybe you did not understand exactly how advanced options works in MtI.

Whenever Mti Use Adv Opt is off, bidir and mlt settings from the UI are discarded. The following only are then written to igs,

Code: Select all

<renderer_settings>
	<width>
	<height>
	<super_sample_factor>
	<logging>
	<image_save_period>
	<display_period>
	<save_igi>
	<save_tonemapped_exr>
	<save_untonemapped_exr>
</renderer_settings>

output truncated horizontally for readability

... and Indigo will use the default renderer values defined in inifile.txt (Idigo app dir, metropolis off, bidir on), every time advanced options are off.

So if you really want to test rendering combinations, leave advanced options ON in MtI and tick/untick mlt/bidir in ui.

I can explain again, or further if you need.

By the way the third shot is a rare thing o.O It's mlt only.